When safety is stripped, this defense poisons the loot: decoys that look right but are wrong
Open-weight safety alignment can be removed in minutes via abliteration, and no release-time defense prevents it. "Fool's Gold" concedes the refusal strip and poisons its payoff: attacked models confidently emit fluent but falsified answers to hazardous requests, cutting reconstructible procedures from 0.58-0.96 down to as low as 0.083 with no label-free way to tell. Tested across seven models (9B-122B), most pass an efficacy gate within benign-behavior and capability budgets.